RESEARCH ARTICLE

PRD-2 directly regulates casein I and counteracts nonsense-mediated decay in the Neurospora circadian Christina M Kelliher1, Randy Lambreghts1, Qijun Xiang1, Christopher L Baker1,2, Jennifer J Loros3, Jay C Dunlap1*

1Department of Molecular & Systems Biology, Geisel School of Medicine at Dartmouth, Hanover, United States; 2The Jackson Laboratory, Bar Harbor, United States; 3Department of Biochemistry & Cell Biology, Geisel School of Medicine at Dartmouth, Hanover, United States

Abstract Circadian clocks in fungi and animals are driven by a functionally conserved –translation loop. In , negative feedback is executed by a complex of Frequency (FRQ), FRQ-interacting RNA (FRH), and casein kinase I (CKI), which inhibits the activity of the clock’s positive arm, the White Collar Complex (WCC). Here, we show that the prd-2 (-2) , whose is characterized by recessive inheritance of a long 26 hr period phenotype, encodes an RNA-binding that stabilizes the ck-1a transcript, resulting in CKI protein levels sufficient for normal rhythmicity. Moreover, by examining the molecular basis for the short circadian period of upf-1prd-6 mutants, we uncovered a strong influence of the Nonsense-Mediated Decay pathway on CKI levels. The finding that circadian period defects in two classically derived Neurospora clock mutants each arise from disruption of ck-1a regulation is consistent with circadian period being exquisitely sensitive to levels of casein kinase I.

*For correspondence: [email protected] Introduction The Neurospora circadian oscillator is a transcription–translation feedback loop that is positively reg- Competing interests: The ulated by the White Collar Complex (WCC) transcription factors, which drive expression of the nega- authors declare that no tive arm component Frequency (FRQ). In this way the fungal core circadian oscillator shares a competing interests exist. common regulatory architecture with the mammalian core clock. In Neurospora, the circadian nega- Funding: See page 17 tive arm complex is composed of FRQ and FRQ-interacting RNA helicase (FRH), which together Received: 14 October 2020 bring casein kinase I (CKI) to promote phosphorylation of WCC on key phospho-sites to inhibit its Accepted: 08 December 2020 activity (Wang et al., 2019). FRQ is extensively regulated transcriptionally, translationally, and post- Published: 09 December 2020 translationally over the circadian day leading ultimately to its inactivation (reviewed in: Hurley et al., 2016). Reviewing editor: Detlef Indeed, in both animals and fungi, the negative arm components are regulated at the RNA and Weigel, Max Planck Institute for Developmental Biology, protein levels to maintain circadian phase and period, and many of the molecular details of this regu- Germany lation, the focus of this paper, are conserved. Negative arm components FRQ and PER are regulated by anti-sense transcription (Koike et al., 2012; Kramer et al., 2003), by thermally regulated splicing Copyright Kelliher et al. This (Colot et al., 2005; Majercak et al., 1999), and display characteristics of intrinsically disordered pro- article is distributed under the teins (Pelham et al., 2020). Another highly conserved feature of fungal, insect, and mammalian neg- terms of the Creative Commons Attribution License, which ative arm components is progressive phosphorylation leading to their inactivation (Baker et al., permits unrestricted use and 2009; Ode et al., 2017; Vanselow et al., 2006) (reviewed in: Dunlap and Loros, 2018). Taken redistribution provided that the together, FRQ, PERs, and CRYs are tightly regulated and underlying mechanisms are often con- original author and source are served between clock models despite evolutionary sequence divergence of these negative arm credited. components.

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 1 of 22 Research article Genetics and Genomics In contrast, less is known about the mechanisms regulating expression of the other essential member of the negative arm complex, CKI, orthologs of which are highly conserved in sequence and in function across eukaryotic clocks. CKI forms a stable complex as FRQ–FRH–CKIa in Neuros- pora (Baker et al., 2009; Go¨rl et al., 2001), as PER- (DBT) in flies (Kloss et al., 2001), and as a multi-protein complex of PER-CRY-CKId in mouse (Aryal et al., 2017). Fungal CKI phos- phorylates both FRQ and WCC (He et al., 2006). Insect DBT and mammalian CKId/e are key regula- tors of the PER2 phospho-switch, differentially phosphorylating two regions that control PER2 turnover (Top et al., 2018; Zhou et al., 2015). Thus, CKI phosphorylations contribute to feedback loop closure in all species. FRQ–CKI binding strength is a key regulator of period length and an important oscillator variable first described in Neurospora (Liu et al., 2019). CKI abundance is not rhythmic in any species described to date (Go¨rl et al., 2001; Kloss et al., 2001), but preliminary evi- dence suggests that its expression levels are tightly controlled to keep the clock on time, just like FRQ/PER/CRY. In mammals, CKI knockdown or knockout significantly lengthens period (Isojima et al., 2009; Lee et al., 2009; Tsuchiya et al., 2016), and CKId levels are negatively regu- lated by m6A methylation (Fustin et al., 2018). In Neurospora, decreasing the amounts of the casein kinase I (ck-1a) transcript using a regulatable leads to long period defects up to ~30 hr (Mehra et al., 2009). CKI has a conserved C-terminal domain involved in and inhibition of kinase activity (Gietzen and Virshup, 1999; Guo et al., 2019). Fungal mutants lacking this CKI C-terminal inhibitory domain have hyperactive kinase activity (Querfurth et al., 2007). Across clock models, the circadian period is sensitive to CKI abundance and activity due to its impor- tance in circadian feedback loop closure. Our modern understanding of the was founded on genetic screens and characteri- zation of mutants with circadian defects (Feldman and Hoyle, 1973; Konopka and Benzer, 1971; Ralph and Menaker, 1988). The fungal clock model Neurospora crassa has been a top producer of relevant circadian mutants due to its genetic tractability, ease of circadian readout, and functional conservation with the animal circadian clock (reviewed in: Loros, 2020). Forward genetic screens used the ras-1bd mutant background (which forms distinct bands of conidiophores once per subjec- tive night) in race tube (RT) assays to identify key players in the circadian clock (Belden et al., 2007; Feldman and Hoyle, 1973; Sargent et al., 1966). Genetic epistasis among the period , and in some cases, genetic mapping of was also performed using N. crassa (Feldman and Hoyle, 1976; Gardner and Feldman, 1981; Morgan and Feldman, 2001). The period (prd) mutants in Neurospora are distinct from the Drosophila gene period (per) (Konopka and Benzer, 1971) and its mammalian orthologs. All but one of the extant period genes in Neurospora have been cloned, and their identities have expanded our knowledge of core-clock modifying processes. prd-4 (period-4), encoding checkpoint kinase 2 (Chk2), links the clock to cell- progression (Pregueiro et al., 2006). prd-3 (period-3), encoding casein kinase II (CKII), implicated direct phosphorylation of core clock as central to temperature compensation (Mehra et al., 2009). prd-1 (period-1) encodes an essential RNA-heli- case that regulates the core clock under high nutrient environments (Emerson et al., 2015). prd-6 (period-6, hereafter referred to as upf1prd-6) encodes the core UPF1 subunit of the Nonsense-Medi- ated Decay (NMD) complex (Compton, 2003), although its circadian role remains cryptic. Among the available prd genes, only prd-2 (period-2) remains uncharacterized. We have mapped the prd-2 mutation to NCU01019 using whole genome sequencing, and discov- ered its molecular identity; however, attributing its long period mutant phenotype to molecular func- tion has remained elusive (Lambreghts, 2012). Equipped with the identity of PRD-2, we then followed up on the observation that the upf1prd-6 short period phenotype is completely epistatic to the prd-2 mutant’s long period (Morgan and Feldman, 1997; Morgan and Feldman, 2001). We find that UPF1PRD-6 and PRD-2 use distinct mechanisms to play opposing roles in regulating levels of the casein kinase I transcript in Neurospora, thus rationalizing the circadian actions of the two clock mutants whose roles in the clock were not understood. PRD-2 stabilizes the ck-1a mRNA transcript, and the clock-relevant domains and biochemical evaluation of the PRD-2 protein indicate that it acts as an RNA-binding protein. We genetically rescue the long period phenotype of prd-2 mutants by expressing a hyperactive CKI and by titrating up ck-1a mRNA levels using a regulatable pro- moter. The endogenous ck-1a transcript has a strikingly long 3’-UTR, indicating that its mRNA could be subject to NMD during a normal circadian day. We confirm that upf1prd-6 mutants have elevated levels of ck-1a in the absence of NMD, and further rescue the short period defect of upf1prd-6

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 2 of 22 Research article Genetics and Genomics mutants by titrating down ck-1a mRNA levels using an inducible promoter. Taken together, a unify- ing model emerges to explain the action of diverse period mutants, where the casein kinase I tran- script is subject to complex regulation by NMD and an RNA-binding protein, PRD-2, to control its and maintain a normal circadian period.

Results An interstitial inversion identifies prd-2 Genetic mapping and preliminary analyses identified prd-2 as a recessive mutant with an abnormally long ~26 hr period length that mapped to the right arm of LG V (Morgan and Feldman, 1997; Morgan and Feldman, 2001). Genetic fine structure mapping using selectable markers flanking prd- 2, in preparation for an anticipated walk, revealed an extensive region of suppressed recombination in the region of the gene, consistent with the existence of a chromosome inversion (Lambreghts, 2012). PCR data consistent with this prompted whole genome sequencing that revealed a 322 kb inversion on chromosome V (Lambreghts, 2012) in the original isolate strain here- after referred to as prd-2INV. The left breakpoint of the inversion occurs in the 5’-UTR of NCU03775, and its upstream regulatory sequences are displaced in the prd-2INV mutant. However, a knockout of NCU03775 (FGSC12475) has a wild-type circadian period length, unlike the long period prd-2INV mutant (Figure 1—figure supplement 1). The next closest gene upstream of the left inversion is NCU03771, but its transcription start site (TSS) is >7 kb away. The right breakpoint of the inversion occurs in the 5’-UTR of NCU01019, disrupting 333 bases of its 5’-UTR and its entire promoter region (Figure 1A and B). A knockout of NCU01019 has a 26 hr long period, matching the prd-2INV long period phenotype (Figure 1C). The prd-2INV mutant has drastically reduced levels of NCU01019 gene expression in constant light conditions and in the subjective evening of a circadian free run (Figure 1D), suggesting that the inversion completely disrupts the NCU01019 promoter and TSS. Placing NCU01019 under the nutrient-responsive qa-2 promoter, we find that the long period length À occurs at very low gene expression levels using 10 6 M quinic acid induction (Figure 1E). Finally, ectopic expression of NCU01019 at the csr-1 in the prd-2INV background rescues the long period phenotype (Figure 1F). We conclude that PRD-2 is encoded by NCU01019. We mapped the clock-relevant domains of the PRD-2 protein (Figure 2A), finding that both an SUZ domain and the proline-rich C-terminus of PRD-2 are required for a normal clock period. This result was confirmed in two separate genetic backgrounds either by replacing the endogenous locus with domain deletion mutants (Figure 2B) or by ectopic expression of domain mutants at the csr-1 locus in a Dprd-2 background (Figure 2C; Supplementary file 1). The SUZ domain family can bind RNA directly in vitro (Song et al., 2008), but curiously PRD-2’s adjacent R3H domain, which is better characterized in the literature as a conserved RNA-binding domain, is dispensable for clock function. The C-terminus of PRD-2 is predicted to be highly disordered, and finer mapping of this region showed that neither a glutamine/proline-rich domain (amino acids 525–612, 21% Gln, 26% Pro) nor a domain conserved across fungal orthologs (amino acids 625–682, 21% Pro) were required for normal clock function (Figure 2C). The remainder of the C-terminus (amino acids 495–524, 28% Pro; 683– 790, 24% Pro) contains a clock-relevant region of PRD-2 based on deletion analyses. Further, PRD-2 SUZ domain and C-terminal deletion mutants are expressed at the protein level, indicating that clock defects must be due to the absent domain (Figure 2—figure supplement 1A). PRD-2 is exclusively localized to the cytoplasm based on biochemical evaluation, and this localization does not change as a function of time of day (Figure 2D). NCU01019 RNA expression is not induced by light (Wu et al., 2014) nor rhythmically expressed over circadian time (Hurley et al., 2014). NCU01019 protein is abundant and shows weak rhythms (Hurley et al., 2018; Figure 2—figure supplement 1B), which suggests that PRD-2 oscillations are driven post-transcriptionally to peak in the early subjective morning, prior to the peak in the frq tran- script (Aronson et al., 1994). Rhythms in PRD-2 protein expression were confirmed using a lucifer- ase translational fusion (Figure 2—figure supplement 1C), which peaked during the circadian day. prd-2INV and DNCU01019 have a slight growth defect (Figure 1C) and are less fertile than wild type as the female partner in a sexual cross (data not shown). Temperature and nutritional compensation of DNCU01019 alone are normal (Figure 2—figure supplement 2), which was expected given the normal TC profile of the prd-2INV mutant (Gardner and Feldman, 1981). PRD-2 (XP_961631.1) is

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 3 of 22 Research article Genetics and Genomics

Figure 1. The prd-2 phenotype derives from reduced expression of NCU01019. Whole genome sequencing identified a 322,386 bp inversion on linkage group V in the original prd-2 mutant strain (Lambreghts, 2012). The inversion breakpoints disrupt two loci, NCU03775 and NCU01019, depicted in cartoon form (A). Sanger sequencing confirms the DNA sequence of the left and right breakpoints, and the corresponding NC12 genome coordinates are shown at each arrowhead (B). Circadian period length was determined by race tube (RT) assay for ras-1bd controls, targeted deletion of the NCU01019 locus, and the classically derived prd-2INV mutant. The DNCU01019 mutant has a long period and slow growth defect similar to prd-2INV (C). NCU01019 RNA expression levels are detectable by RT-qPCR in the prd-2INV mutant but are drastically reduced compared to ras-1bd controls grown in constant light (LL) or at subjective dusk (CT12) during a circadian free run (D). After replacing the endogenous promoter of NCU01019 with the À À À inducible qa-2 promoter, addition of high levels of quinic acid (10 2 to 10 3 M) led to a normal circadian period by RT assay (10 2 M t = 21.5 ± 0.2 hr; À À À À 10 3 M t = 21.6 ± 0.3 hr; 10 4 M t = 21.5 ± 0.2 hr). Lower levels of QA inducer led to a long circadian period (10 5 M t = 22.8 ± 0.3 hr; 10 6 M À t = 24.3 ± 0.2 hr; 0 QA t = 24.6 ± 0.3 hr) due to reduced NCU01019 expression. Asterisks (**) indicate p<1 Â 10 10 by Student’s t-test compared to Figure 1 continued on next page

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 4 of 22 Research article Genetics and Genomics

Figure 1 continued À 10 2 M QA RT results (E). The entire NCU01019 locus (plus 951 bases of its upstream promoter sequence) was fused in-frame with codon-optimized . Ectopic expression of this NCU01019-luc construct in the prd-2INV background rescues the long period phenotype by RT assay (F). The online version of this article includes the following figure supplement(s) for figure 1: Figure supplement 1. NCU03775 knockout has a normal circadian period and does not explain the prd-2INV phenotype.

well conserved among Ascomycota fungi as noted by BLASTp scores (

PRD-2 regulates CKI levels To identify the putative mRNA targets of PRD-2, we performed total RNA-sequencing on triplicate samples of Dprd-2 versus control grown in constant light at 25˚C. Hundreds of genes are affected by loss of PRD-2, but we did not identify a consensus functional category or sequence motif(s) for the putative PRD-2 regulon (Figure 3—figure supplement 1). Given the pleotropic phenotypes of Dprd- 2, we posit that PRD-2 plays multiple roles in the cell, including regulation of carbohydrate and sec- ondary metabolism. Focusing specifically on core clock genes, we found that ck-1a, frq, wc-2, ckb-1 (regulatory beta subunit of CKII), and frh were significantly altered in the absence of PRD-2 (Figure 3A). Pursuing the top two hits, we found that the CKI transcript was dramatically less stable in Dprd-2 (Figure 3B), while frq mRNA stability was not significantly altered (Figure 3—figure sup- plement 2). To demonstrate that PRD-2 binds the ck-1a transcript in vivo, we used RNA immunopre- cipitation after UV cross-linking (CLIP). The Pumilio family RNA-binding protein PUF4 (NCU16560) was previously shown to bind in the 3’-UTR of cbp3 (NCU00057), mrp-1 (NCU07386), and other tar- get genes identified by HITS-CLIP high-throughput sequencing (Wilinski et al., 2017). C-terminally tagged of PRD-2, PUF4, and an untagged negative control strain were used to immunopre- cipitate cross-linked RNAs (Materials and methods). As expected, cbp3 and mrp-1 positive controls were significantly enriched in the PUF4 CLIP sample compared to the negative IP (Figure 3C). ck-1a is also enriched in the PRD-2 CLIP sample, demonstrating that the CKI transcript is a direct target of the PRD-2 protein (Figure 3C). Hypothesizing that the clock-relevant target of PRD-2 could be CKI, we used two genetic approaches to manipulate CKI activity in an attempt to rescue the Dprd-2 long period phenotype. First, we placed the ck-1a gene under the control of the quinic acid inducible promoter (Mehra et al., 2009) and crossed this construct into the Dprd-2 background. We found that increas- À À ing expression of ck-1a using high levels (10 1 to 10 2 M) of QA partially rescued the Dprd-2 long period phenotype (Figure 4A). We also noticed a synergistic poor growth defect in the double À mutant at 10 4 M QA, consistent with low levels of ck-1a (an essential gene in Neurospora: Go¨rl et al., 2001; He et al., 2006). There are two explanations for the lack of full rescue to periods À1 shorter than 25 hr in the Pqa-2-ck-1a Dprd-2 double mutant: (1) even at saturating 10 M QA induc- tion, the qa-2 promoter may not reach endogenous levels of ck-1a achieved under its native pro- moter, and/or (2) because PRD-2 acts directly as an RNA-binding protein for CKI transcripts, simply increasing levels of ck-1a RNA cannot fully rescue PRD-2’s role in stabilizing or positioning CKI tran- scripts in the cytoplasm. Next, we turned to a previously described fungal CKI constitutively active allele, CKI Q299STOP (Querfurth et al., 2007), reasoning that we might be able to rescue low ck-1a levels in Dprd-2 by genetically increasing CKI kinase activity. We replaced endogenous CKI with a CKISHORT allele, which expresses only the shortest ck-1a isoform (361 amino acids). CKISHORT lacks 23 amino acids in the C-terminal tail of the full length isoform that are normally subject to autophosphorylation leading to kinase inhibition. This CKISHORT allele also carries an in-frame C-terminal HA3 tag and selectable marker, which displace the endogenous 3’-UTR of ck-1a. The CKISHORT mutant has a short period phenotype (~17 hr), presumably due to hyperactive kinase activity and rapid feedback loop closure (Liu et al., 2019). Significantly, the CKISHORT mutation is completely epistatic to Dprd-2 (Figure 4B), indicating that CKI is the clock-relevant target of PRD-2.

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 5 of 22 Research article Genetics and Genomics

Figure 2. Clock-relevant protein domains and localization of PRD-2 suggest and RNA-binding function. PRD-2 has tandemly arrayed R3H and SUZ domains associated with RNA binding proteins, and its C-terminal region is highly enriched for proline (P) and glutamine (Q). The cartoon of PRD-2 protein lists relevant coordinates (A). The native NCU01019 locus was replaced with single domain deletion mutants, and 96-well plate luciferase assays were used to measure the circadian period length in triplicate wells per biological replicate experiment. A wild-type clock period was recovered in ras-1bd controls and the prd-2DR3H mutant, while Dprd-2, prd-2DSUZ, and prd-2DC-terminus had long period phenotypes (B). Independently constructed strains targeted domain deletion mutants to the csr-1 locus in a Dprd-2 background (Supplementary file 1), and mutant period lengths were determined by race tube assay. Period lengths (±1 SD) show that the clock-relevant domains of PRD-2 are the SUZ domain and the C-terminus (C). Total (T), Nuclear (N), and Cytosolic (C) fractions were prepared over a circadian time course (N = 1 per time point). g-Tubulin (NCU03954) was used as a control for cytoplasmic localization and histone H3 (NCU01635) for nuclear localization. PRD-2 tagged with a C-terminal V5 epitope tag is localized to the cytoplasm throughout the circadian cycle (D). The online version of this article includes the following figure supplement(s) for figure 2: Figure supplement 1. PRD-2 protein levels are slightly rhythmic and are detectable in protein domain deletion mutants. Figure supplement 2. Temperature and nutritional compensation are normal in DNCU01019.

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 6 of 22 Research article Genetics and Genomics

Figure 3. The core clock target of PRD-2 is the casein kinase I transcript. Control and Dprd-2 cultures were grown in the light at 25˚C in Bird medium for 48 hr prior to RNA isolation. Expression levels for core clock genes were measured by RNA-sequencing (N = 3 biological replicates per strain), and À5 log2-transformed FPKM values are shown. Asterisks indicate p<0.05 (*) or p<5 Â 10 (***) by Student’s t-test compared to control levels. The ck-1a transcript is >1.5Â less abundant in Dprd-2 (A). ck-1a mRNA degradation kinetics were examined by Northern blot in a time course after treatment with thiolutin (THL) at approximately CT1 (N = 2 biological replicates). RNA levels were quantified using ImageJ, natural log transformed, fit with a linear model (glm in R, Gaussian family defaults), and half-life was calculated assuming first order decay kinetics (ln(2)/slope). Shaded areas around the linear fit represent 95% confidence intervals on the slope. The ck-1a transcript is 3Â less stable in Dprd-2 (B). The PUF4 (NCU16560) RNA-binding protein pulls down known target transcripts cbp3 (NCU00057) and mrp-1 (NCU07386) by RT-qPCR (N = 3 biological replicates). PRD-2 CLIP samples were processed in parallel with PUF4 positive controls, and PRD-2 binds the ck-1a transcript in vivo (C). The online version of this article includes the following figure supplement(s) for figure 3: Figure supplement 1. Hundreds of genes have altered expression levels in the Dprd-2 mutant but a common pathway or sequence motif was not detected. Figure supplement 2. Loss of prd-2 has little effect on stability of the frq transcript.

NMD impacts the clock by regulating CKI levels NMD in Neurospora crassa is triggered by various mRNA signatures. Open reading frames in 5’- UTRs that produce short peptides (5’-uORFs) can trigger NMD in a mechanism that does not require the Exon Junction Complex (EJC; Zhang and Sachs, 2015). The frq transcript has six such uORFs (Colot et al., 2005; Diernfellner et al., 2005) and could be a bona fide NMD target because its splicing is disrupted in the absence of NMD (Wu et al., 2017). Transcripts containing long 3’-UTRs are also subject to NMD regulation. In addition, transcripts with (s) near a STOP codon and/or

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 7 of 22 Research article Genetics and Genomics

Figure 4. Genetically increasing casein kinase I (CKI) levels or activity rescues the Dprd-2 long period phenotype. Representative race tubes (RTs) from bd bd ras-1 Pqa-2-ck-1a single (pink) and ras-1 Pqa-2-ck-1a Dprd-2 double (yellow) mutants are shown with growth using the indicated concentrations of quinic acid (QA) to drive expression of ck-1a. All results are shown in a scatterplot, where each dot represents one RT’s free running period length. ras- 1bd controls (black) had an average period of 22.5 ± 0.5 hr (N = 12), and period length was not significantly affected by QA concentration (ANOVA Figure 4 continued on next page

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 8 of 22 Research article Genetics and Genomics

Figure 4 continued p=0.297). ras-1bd Dprd-2 controls (blue) had an average period of 25.4 ± 0.4 hr (N = 10), and period length was not significantly affected by QA bd concentration (ANOVA p=0.093). Period length of ras-1 Pqa-2-ck-1a single mutants (pink) was significantly altered across QA levels (ANOVA À6 À1 bd p=3.6 Â 10 ), and the average period at 10 M QA was 24.3 ± 0.5 hr (N = 4). Period length of ras-1 Pqa-2-ck-1a Dprd-2 double mutants (yellow) was À À also significantly affected by QA levels (ANOVA p=8.1 Â 10 8), and the average period at 10 1 M QA was 25.4 ± 0.4 hr (N = 4). The double mutant period length was not genetically additive at high levels of QA induction (A). A hyperactive CKI allele was constructed by expressing the shortest isoform only (CKISHORT). 96-well plate luciferase assays were used to measure the circadian period length. Traces represent the average of three technical replicates across four biological replicate experiments for: ras-1bd controls (gray, t = 21.7 ± 0.3 hr), ras-1bd Dprd-2 (blue, t = 25.7 ± 0.6 hr), ras- 1bd CKISHORT (pink, t = 17.4 ± 0.3 hr), and ras-1bd CKISHORT Dprd-2 double mutants (yellow, t = 18.2 ± 0.3). CKISHORT is completely epistatic to Dprd-2 in double mutants (B).

with intron(s) in the 3’-UTR can be degraded by NMD after recruitment of the UPF1/2/3 complex by the EJC in a pioneering round of translation (Zhang and Sachs, 2015). Since the observation by Compton, 2003 that the short period mutant prd-6 identified the UPF1 core subunit of the NMD pathway, the clock-relevant target(s) of NMD has been an object of conjec- ture and active research. Because loss of NMD reduces the amount of the transcript encoding the short-FRQ (Wu et al., 2017), and strains making only short-FRQ have slightly length- ened periods (Liu et al., 1997), Wu et al., 2017 recently speculated that the short period of the upf1prd-6 mutant might be explained by effects of NMD on FRQ. However, strains expressing only long-FRQ display an essentially wild-type period length (Colot et al., 2005; Liu et al., 1997), not a short period phenotype like upf1prd-6; this finding is not consistent with FRQ being the only or even principal clock-relevant target of NMD, leaving unresolved the role of NMD in the clock. To tackle this puzzle, we returned to classical genetic epistasis experiments and confirmed the observation that upf1prd-6 is completely epistatic to prd-2INV (Morgan and Feldman, 2001), going on to show that in fact each of the individual NMD subunit knockouts, Dupf2 and Dupf3 as well as Dupf1prd-6, is epistatic to the Dprd-2 long period phenotype (Figure 5A). Previous work had profiled the transcriptome of Dupf1prd-6 compared to a control (Wu et al., 2017); we re-processed this RNA- seq data and found, exactly as in Dprd-2, that ck-1a was the most affected core clock gene in Dupf1prd-6 (Figure 5B). The ck-1a transcript has an intron located 70 nt away from its longest iso- form’s STOP codon, and its 3’-UTR is, remarkably, among the 100 longest annotated UTRs in the entire Neurospora transcriptome (Figure 5C). NMD targeting to long 3’-UTR transcripts like ck-1a is thought to occur independently of the EJC and nuclear cap-binding complex (CBC) in Neurospora crassa (Zhang and Sachs, 2015). We used the knockout mutant Dcbp80 (NCU04187) to confirm that Neurospora CBC is not required for a normal circadian clock and that the long period length of Dprd-2 is unchanged in the Dcbp80 background (Figure 5—figure supplement 1). Thus, ck-1a is a strong candidate for NMD-mediated degradation via its long 3’-UTR, not dependent on EJC and CBC components. We hypothesized that CKI is overexpressed in the absence of NMD (Figure 5B), leading to faster feedback loop closure and a short circadian period. To genetically control ck-1a levels, we crossed prd-6 the regulatable Pqa-2-ck-1a allele into the Dupf1 background and confirmed our hypothesis by À finding that at low levels of inducer (10 5 M QA), decreased levels of ck-1a transcript revert the short period length of Dupf1prd-6 to control period lengths (Figure 5D). Further, protein levels of CKI in À the Dupf1prd-6 background are reduced to control levels at 10 5 M QA (Figure 5E), which explains the period rescue phenotype. CKI protein is two to three times more abundant in Dupf1prd-6 and in Dprd-2 Dupf1prd-6 (Figure 5F), matching its overexpression in the Dupf1prd-6 transcriptome (Figure 5B). CKI protein is 3Â reduced in Dprd-2 (Figure 5F), also correlating with its reduced mRNA expression and stability (Figure 3). We conclude that CKI is also the clock-relevant target of UPF1PRD-6, placing NMD, PRD-2, and CKI in the same genetic epistasis pathway.

Discussion By uncovering the identity and mode of action of PRD-2 and exploring the mechanism of two classi- cal period mutants, prd-2 and upf1prd-6, we found a common basis in regulation of CKI levels, which are under tight control in the Neurospora clock (Figure 6). That the mechanistic basis of action of two independently derived non-targeted clock mutants centers on regulation of the activity of a

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 9 of 22 Research article Genetics and Genomics

Figure 5. Nonsense-mediated decay (NMD) negatively regulates casein kinase I (CKI) levels via UPF1PRD-6, establishing a basis for the upf1prd-6 prd-2 genetic epistasis on circadian period length. 96-well plate luciferase assays were used to measure the circadian period length in triplicate wells per three biological replicate experiments for: ras-1bd controls (black, t = 21.5 ± 0.3 hr), ras-1bd Dprd-2 (blue, t = 25.5 ± 0.4 hr); ras-1bd Dupf1prd-6 (purple, t = 18.1 ± 0.2 hr), ras-1bd Dupf1prd-6Dprd-2 double mutants (yellow, t = 19.4 ± 0.7 hr); ras-1bd Dupf2 (purple, t = 18.5 ± 0.5 hr), ras-1bd Dupf2 Dprd-2 Figure 5 continued on next page

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 10 of 22 Research article Genetics and Genomics

Figure 5 continued double mutants (yellow, t = 18.1 ± 0.3 hr); ras-1bd Dupf3 (purple, t = 19.8 ± 0.3 hr), ras-1bd Dupf3 Dprd-2 double mutants (yellow, t = 20.1 ± 0.2 hr). Each individual NMD subunit knockout is epistatic to the Dprd-2 long period phenotype (A). Raw RNA-seq data from a previous study (Wu et al., 2017) were prd-6 analyzed using the same pipeline as data from Figure 3A (see Materials and methods). Control and Dupf1 gene expression levels (log2- transformed) are shown for core clock genes. The ck-1a transcript is >2Â more abundant in Dupf1prd-6 (B). 3’-UTR lengths from 7793 genes were mined from the N. crassa OR74A genome annotation (FungiDB version 45, accessed on 10/25/2019), and plotted as a histogram. The arrow marks the 3’-UTR bd of ck-1a, which is 1739 bp and within the top 100 longest annotated UTRs in the entire genome (C). Representative race tubes (RTs) from ras-1 Pqa-2- bd prd-6 ck-1a single (pink) and ras-1 Pqa-2-ck-1a Dupf1 double (yellow) mutants are shown at the indicated concentrations of quinic acid to drive expression of ck-1a. All results are shown in a scatterplot, where each dot represents one RT’s free running period length. ras-1bd controls (black) had an average period of 22.4 ± 0.4 hr (N = 20), and period length was not significantly affected by QA concentration (ANOVA p=0.605). ras-1bd Dupf1prd-6 controls (purple) had an average period of 17.5 ± 0.6 hr (N = 16), and period length was not significantly affected by QA concentration (ANOVA bd À8 p=0.362). Period length of ras-1 Pqa-2-ck-1a single mutants (pink) was significantly altered across QA levels (ANOVA p=2.9Â10 ), and the average À5 bd prd-6 period at 10 M QA was 27.6 ± 0.8 hr (N = 8). Period length of ras-1 Pqa-2-ck-1a Dupf1 double mutants (yellow) was also significantly affected by À À QA levels (ANOVA p=9.4Â10 12), and the average period at 10 5 M QA was 24.7 ± 0.9 hr (N = 8). Thus, the double mutant period length was not genetically additive at low levels of QA induction, and the short period phenotype of Dupf1prd-6 is rescued (D). CKI protein levels were measured from the indicated genotypes grown in 0.1% glucose liquid culture medium (LCM) with QA supplemented at the indicated concentrations for 48 hr in constant light. A representative immunoblot of three biological replicates is shown, and replicates are quantified in the bar graph relative to ras-1bd control CKI levels from a 2% glucose LCM culture (E). CKI protein levels were measured from the indicated genotypes grown in 2% glucose LCM for 48 hr in constant light. A representative immunoblot of three biological replicates is shown, and replicates are quantified in the bar graph relative to ras- 1bd control CKI levels (F). CKI protein levels are increased in Dupf1prd-6, decreased in the Dprd-2 mutant, and Dupf1prd-6 is epistatic to Dprd-2 with respect to CKI levels and circadian period length. The online version of this article includes the following figure supplement(s) for figure 5: Figure supplement 1. The cap-binding protein CBP80 (NCU04187) is not required for a normal clock, and does not alter the Dprd-2 long period phenotype, suggesting that ck-1a degradation is controlled by NMD machinery without the Exon Junction Complex and nuclear cap-binding complex. Figure supplement 2. Long untranslated regions (UTRs) are characteristic of casein kinase I gene orthologs across species.

single , CKI, via two distinct mechanisms is noteworthy. prd-2 encodes an RNA-binding pro- tein (Figures 1 and 2) that stabilizes the CKI transcript (Figure 3B). We demonstrate that CKI is the most important core clock target of PRD-2 by rescuing its long period mutant phenotype with a

Figure 6. Counterbalancing regulation of casein kinase I (CKI) provides a unifying genetic model for the action of PRD-2 and UPF1PRD-6 in the circadian oscillator. The NMD complex (UPF1PRD-6, UPF2, and UPF3) targets the frq and ck-1a transcripts for degradation (upstream uORFs in frq; long 3’-UTR in ck-1a). PRD-2 binds to and stabilizes ck-1a transcripts (dashed lines), which could also promote local translation and complex formation for the negative arm of the clock. In the absence of PRD-2, the long period phenotype is due to low CKI levels, and in the absence of NMD, the short period phenotype is due to high CKI levels.

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 11 of 22 Research article Genetics and Genomics hyperactive CKI allele (Figure 4B). The predominantly cytoplasmic localization of PRD-2 (Figure 2D) is consistent with its action in protecting ck-1a transcripts from NMD and rounds out the model. PTBP1, an RNA-binding protein, protects its target transcripts from NMD-mediated degradation by binding in the 3’-UTR and blocking NMD recruitment in mouse (Ge et al., 2016), and future work will determine if PRD-2 functions similar to PTBP1. This work contributes another possible example to the growing literature describing conserved post-transcriptional regulation on core clock messages. Anti-sense transcription at the frq locus pro- duces the qrf transcript, which is required for proper phase control and light responses of the fungal clock (Kramer et al., 2003). The mammalian PER2 anti-sense transcript displays nearly identical dynamics to qrf expression (Koike et al., 2012). Mammalian PER2 sense expression levels are further regulated by microRNA binding sites in its 3’-UTR (Yoo et al., 2017). In a similar manner, frq RNA is directly targeted for turnover by rhythmic exosome activity in the late day (Guo et al., 2009). Splic- ing of the frq transcript is regulated by temperature (Colot et al., 2005), mirroring thermal regula- tion mechanisms in the clocks of Drosophila (Majercak et al., 1999) and Arabidopsis (James et al., 2012). The codons composing the frq transcript are non-optimal, which improves FRQ’s co-transla- tional folding (Zhou et al., 2013), and FRQ’s disordered protein structure is also stabilized by its binding partner FRH (Hurley et al., 2013). Mammalian PER2 is also largely intrinsically disordered, and indeed circadian clock proteins across species have large stretches of intrinsic disorder which are in the early stages of functional characterization (Pelham et al., 2020; Pelham et al., 2018) (reviewed in: Partch, 2020). These data document the complexity of post-transcriptional regulation of clock components, and this study demonstrates that even non-rhythmic clock transcripts such as CKI are under tight regulation that is essential for normal clock function. UPF1PRD-6 and the NMD machinery target ck-1a mRNA for degradation to regulate its expression levels, presumably mediated by the long 3’-UTR of ck-1a transcripts in Neurospora (Figure 5). NMD components are not rhythmic in abundance in the fungal clock (Hurley et al., 2014; Hurley et al., 2018). These data, taken together with the constitutive expression of the CKI mRNA and protein (Baker et al., 2009; Go¨rl et al., 2001; Hurley et al., 2014; Hurley et al., 2018), lead us to predict that NMD regulation of CKI occurs throughout the circadian cycle. To our knowledge the discovery of NMD regulation of CKI represents a wholly novel and potentially important mode of regulation for this pivotal kinase. Future work will investigate whether insect DBT and/or mammalian CKId/e (CSNK1D, CSNK1E) are also targets of NMD. Long UTR length appears to be conserved across CKI orthologs (Figure 5—figure supplement 2). One previous study in Drosophila reported a circadian period defect in a tissue-specific NMD knockdown (Ri et al., 2019), but the behavioral rhythm was lengthened in UPF1-depleted insects unlike the short period defect observed in Neurospora. In mouse, both CKIe and CLOCK display altered splicing patterns in the absence of UPF2 (Weischenfeldt et al., 2012). Most core clock proteins have at least one uORF in mammals (Millius and Ueda, 2017), altogether raising the possibility that multiple core clock genes are regu- lated by NMD. The importance of NMD has already been recognized and investigated in the plant clock, where leads to NMD turnover for four core clock and accessory mRNAs: GRP7, GRP8, TOC1, and ELF3 (reviewed in: Mateos et al., 2018). CKI abundance and alternative isoforms strongly affect circadian period length. Low levels of CKI driven from an inducible promoter lead to long periods approaching 30 hr (Mehra et al., 2009; Figure 4A). In the mammalian clock, decreased CKI expression also significantly lengthens period (Isojima et al., 2009; Lee et al., 2009; Tsuchiya et al., 2016). CKI is rendered hyperactive by remov- ing its conserved C-terminal domain, a domain normally subject to autophosphorylation leading to kinase inhibition (Gietzen and Virshup, 1999; Guo et al., 2019; Querfurth et al., 2007). We gener- ated a CKI mutant expressing only this shortest CKI isoform, finding a 17.5 hr short period pheno- type in the absence of C-terminal autophosphorylation (Figure 4B). Based on prior work, increased CKI activity and/or abundance would be expected to increase FRQ-CKI affinity and lead to faster feedback loop closure (Liu et al., 2019), consistent with the short period phenotype. Curiously, this CKI short isoform is expressed at levels similar to the full length isoform in Neurospora (as well as a third short isoform derived from an alternative splice acceptor event) (Figure 5F), and all isoforms interact with FRQ by immunoprecipitation (Querfurth et al., 2007). Why do natural isoforms arise without the auto-inhibitory C-terminus in Neurospora, and are these regulatory events required to keep the clock on time? Mammalian alternative isoforms CKId1 and CKId2 have different substrate preferences in vitro, which leads to differential phosphorylation of PER2 whereby CKId2

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 12 of 22 Research article Genetics and Genomics phosphorylation significantly stabilizes PER2 (Fustin et al., 2018). Adding further complexity, CKId1 and CKId2 isoform expression patterns appear to be tissue specific and are regulated by m6A RNA modification. Regulation of CKI levels and isoform expression is an important direction for future work in the circadian clock. CKI has a diverse array of functions in eukaryotes and is critically important in human health (reviewed in: Cheong and Virshup, 2011; Vielhaber and Virshup, 2001). CKI overexpression is pathogenic in Alzheimer’s disease in addition to its role in circadian period regulation (Sundaram et al., 2019). Mutation of hPER2 at residue S662 is associated with the human sleep and circadian disorder FASPS (Toh et al., 2001), and CKId/CKIe control the phosphorylation state of this critical site as well as phospho-switch regions dictating PER2 stability (Narasimamurthy et al., 2018; Philpott et al., 2020; Zhou et al., 2015). Significantly, mutation of human CKId itself phenocopies this, also leading to FASPS (Xu et al., 2005). Future work on the regulation of CKI levels and isoform expression will shed light on CKI regulation in the clock, in development, and in human disease.

Materials and methods

Key resources table

Reagent type (species) or Source or Additional resource Designation reference Identifiers information Gene prd-2 FungiDB NCU01019 (Neurospora crassa) Gene upf1prd-6 FungiDB NCU04242 (Neurospora crassa) Gene ck-1a FungiDB NCU00685 (Neurospora crassa) Strain, strain Supplementary This study; Fungal background file 1 Genetics Stock (Neurospora Center (FGSC) crassa) Antibody Anti-V5 (mouse ThermoFisher Cat. # R960-25 (1:3000) monoclonal) Antibody Anti-tubulin Fitzgerald Cat. # 10R-T130a (1:10,000) alpha (mouse monoclonal) Antibody Anti-CKI (rabbit Generous gift from (1:1000) polyclonal) Michael Brunner (University of Heidelberg) Antibody Anti-FLAG M2 Sigma Cat. # M8823 30 ml beads magnetic beads incubated (mouse with 10 mg total protein monoclonal) for UV-CLIP Recombinant c box-luc (plasmid- As described, <500 bp DNA reagent derived construct) PMID:25635104 of the frq promoter driving codon-optimized luciferase; targeted to the csr-1 locus for selection Chemical D-quinic acid Sigma Cat. # 138622 1 M stock solution, pH adjusted to 5.8 with NaOH compound, drug Continued on next page

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 13 of 22 Research article Genetics and Genomics

Continued

Reagent type (species) or Source or Additional resource Designation reference Identifiers information Chemical Allele-In-One Allele Biotechnology Cat. # ABP-PP-MT01500 50 ml reagent compound, Mouse Tail Direct mixed with drug Lysis Buffer Neurospora asexual spores for gDNA isolation Chemical Thiolutin Cayman Chemical Cat. # 11350 Stock compound, solution drug prepared in DMSO Software, Custom R https://github. UTR length analyses algorithm software com/cmk35 from Figure 5—figure supplement 2

Neurospora strains and growth conditions The ras-1bd prd-2INV strains 613–102 (mat A) and 613–43 (mat a) were originally isolated in the Feld- man laboratory (Lewis, 1995). Strains used in this study were derived from the wild-type back- ground (FGSC2489 mat A), ras-1bd background (87–3 mat a or 328–4 mat A), Dmus-51 background (FGSC9718 mat a), or the Fungal Genetics Stock Center (FGSC) knockout collection as indicated (Supplementary file 1). Strains were constructed by transformation or by sexual crosses using stan- dard Neurospora methods (http://www.fgsc.net/Neurospora/NeurosporaProtocolGuide.htm). The ‘c box-luc’ core clock transcriptional reporter was used to assay circadian period length by luciferase (Figure 2B, Figure 4B, Figure 5A, Figure 1—figure supplement 1, and Figure 5—figure supplement 1). In this construct, a codon-optimized firefly luciferase gene is driven by the clock box in the frequency promoter (Gooch et al., 2008; Hurley et al., 2014; Larrondo et al., 2015). The clock reporter construct was targeted to the csr-1 locus and selected on resistance to 5 mg/ml cyclo- sporine A (Sigma # 30024) (Bardiya and Shiu, 2007). Standard race tube (RT) medium was used for all RTs (1Â Vogel’s Salts, 0.1% glucose, 0.17% argi- nine, 1.5% agar, and 50 ng/ml biotin). Where indicated, D-quinic acid (Sigma # 138622) was added from a fresh 1 M stock solution (pH 5.8). Standard 96-well plate medium was used for all camera runs (1Â Vogel’s Salts, 0.03% glucose, 0.05% arginine, 1.5% agar, 50 ng/ml biotin, and 25 mM lucif- erin from GoldBio # 115144-35-9). Liquid cultures were started from fungal plugs as described (Chen et al., 2009; Nakashima, 1981) or from a conidial suspension at 1 Â 105 conidia/ml. Liquid cultures were grown in 2% glucose liquid culture medium (LCM; 1Â Vogel’s Salts, 0.5% arginine w/v) or in 1.8% glucose Bird Medium (Metzenberg, 2004) as indicated. QA induction experiments in liq- uid culture were performed in 0.1% glucose LCM medium with QA supplemented. All the experiments were conducted at 25˚C in constant light unless otherwise indicated. Strains were genotyped by screening for growth on selection medium (5 mg/ml cyclosporine A, 400 mg/ml Ignite, and/or 200–300 mg/ml Hygromycin). PCR genotyping was performed on gDNA extracts from conidia incubated with Allele-In-One Mouse Tail Direct Lysis Buffer (Allele Biotechnol- ogy # ABP-PP-MT01500) according to the manufacturer’s instructions. GreenTaq PCR Master Mix (ThermoFisher # K1082) was used for genotyping. Relevant genotyping primers for key strains are: ras-1bd (mutant): 5’ TGCGCGAGCAGTACATGCGAAT and 5’ CCTGATTTCGCGGACGAGATCGTA 3’; ras-1WT (NCU08823): 5’ GCGCGAGCAGTACATGCGGAC 3’ and 5’ CCTGATTTCGCGGACGAGA TCGTA 3’; prd-2WT (NCU01019): 5’ CACTTCCAGTTATCTCGTCAC 3’ and 5’ CACAACCTTG TTAGGCATCG 3’; Dprd-2::barR (KO mutant): 5’ CACTTCCAGTTATCTCGTCAC 3’ and 5’ GTGCTTG TCTCGATGTAGTG 3’; prd-2INV (left breakpoint): 5’ AGCGAGCTGATATGCCTTGT 3’ and 5’ CGAC TTCCACCACTTCCAGT 3’; prd-2INV (right breakpoint): 5’ TGTTTGTCCGGTGAAGATCA 3’ and 5’ G TCGTGGAATGGGAAGACAT 3’; Dupf1prd-6::hygR (FGSC KO mutant): 5’ CTGCAACCTCGGCCTCCT R 3’ and 5’ CAGGCTCTCGATGAGCTGATG 3’; bar ::Pqa-2-ck-1a (QA inducible CKI): 5’ GTGCTTGTC TCGATGTAGTG 3’ and 5’ GATGTCGCGGTGGATGAACG 3’.

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 14 of 22 Research article Genetics and Genomics RNA stability assays Control and Dprd-2 liquid cultures grown in 1.8% glucose Bird medium were age-matched and circa- dian time (CT) matched to ensure that RNA stability was examined at the same phase of the clock. Control cultures were shifted to constant dark for 12 hr, and Dprd-2 cultures were shifted to dark for 14 hr (~CT1 for 22.5 hr wild-type period and for 26 hr Dprd-2 period; 46 hr total growth). Thiolutin (THL; Cayman Chemical # 11350) was then added to a final concentration of 12 mg/ml to inhibit new RNA synthesis. Samples were collected every 10 min after THL treatment by vacuum filtration and flash frozen in liquid nitrogen. THL has multiple off-target effects in addition to inhibiting transcrip- tion (Lauinger et al., 2017). For this reason, frq mRNA degradation kinetics were also examined with an alternative protocol. Light-grown, age-matched liquid Bird cultures of wild-type and Dprd-2 were shifted into the dark and sampled every 10 min to measure frq turnover; transcription of frq ceases immediately on transfer to darkness (Heintzen et al., 2001; Tan et al., 2004). All tissue manipulation in the dark was performed under dim red lights, which do not reset the Neurospora clock (Chen et al., 2009).

RNA isolation and detection Frozen Neurospora tissue was ground in liquid nitrogen with a mortar and pestle. Total RNA was extracted with TRIzol (Invitrogen # 15596026) and processed as described (Chen et al., 2009). RNA samples were prepared for RT-qPCR, northern blotting, RNA-sequencing, or stored at À80˚C. For RT-qPCR, cDNA was synthesized using the SuperScript III First-Strand synthesis kit (Invitrogen # 18080–051). RT-qPCR was performed using SYBR green master mix (Qiagen # 204054) and a Ste-

pOne Plus Real-Time PCR System (Applied Biosystems). Ct values were determined using StepOne

software (Life Technologies) and normalized to the actin gene (DCt). The DDCt method was used to determine mRNA levels relative to a reference time point. Relevant RT-qPCR primer sequences are: prd-2 (NCU01019): 5’ GGGCAACGACGTCAAACTAT 3’ and 5’ TGCGTGTACATCACTCTGGA 3’, and actin (NCU04173): 5’ GGCCGTGATCTTACCGACTA 3’ and 5’ TCTCCTTGATGTCACGAACG 3’. Northern probes were first synthesized using the PCR DIG Probe Synthesis Kit (Roche # 11 636 090 910). The 512 bp frq probe was amplified from wild-type Neurospora genomic DNA with pri- mers: 5’ CTCTGCCTCCTCGCAGTCA 3’ and 5’ CGAGGATGAGACGTCCTCCATCGAAC 3’. The 518 bp ck-1a probe was amplified with primers: 5’ CCATGCCAAGTCGTTCATCC 3’ and 5’ CGGTCCAG TCAAAGACGTAGTC 3’. Total RNA samples were prepared according to the NorthernMax-Gly Kit instructions (Invitrogen # AM1946). Equal amounts of total RNA (5–10 mg) were loaded per lane of a 0.8–1% w/v agarose gel. rRNA bands were visualized prior to transfer to validate RNA integrity. Transfer was completed as described in the NorthernMax-Gly instructions onto a nucleic acid Amer- sham Hybond-N+ membrane (GE # RPN303B). Transferred RNA was cross-linked to the membrane using a Stratalinker UV Crosslinker. The membrane was blocked and then incubated overnight at 42˚ C in hybridization buffer plus the corresponding DIG probe. After washing with NorthernMax-Gly Kit reagents, subsequent washes were performed using the DIG Wash and Block Buffer Set (Roche # 11 585 762 001). Anti-Digoxigenin-AP Fab fragments were used at 1:10,000. Chemiluminescent detec- tion of anti-DIG was performed using CDP-Star reagents from the DIG Northern Starter Kit (Roche # 12 039 672 910). Densitometry was performed in ImageJ. Total RNA was submitted to Novogene for stranded polyA+ library preparation and sequencing. 150 bp paired-end (PE) read libraries were prepared, multiplexed, and sequenced in accordance with standard Illumina HiSeq protocols. 24.8 ± 1.7 million reads were obtained for each sample. Raw FASTQ files were aligned to the Neurospora crassa OR74A NC12 genome (accessed September 28, 2017, via the Broad Institute: ftp://ftp.broadinstitute.org/pub/annotation/fungi/neurospora_crassa/ assembly/) using STAR (Dobin et al., 2013). On average, 97.6 ± 0.3% of the reads mapped uniquely to the NC12 genome. Aligned reads were assembled into transcripts, quantified, and normalized using Cufflinks2 (Trapnell et al., 2013). Triplicate control and Dprd-2 samples were normalized together with CuffNorm, and the resulting FPKM output was used in the analyses presented. RNA- sequencing data have been submitted to the NCBI Gene Expression Omnibus (GEO; https://www. ncbi.nlm.nih.gov/geo/) under accession number GSE155999.

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 15 of 22 Research article Genetics and Genomics CLIP assay CLIP was performed using PUF4 (NCU16560) as a positive control RNA-binding protein from Wilinski et al., 2017, with modifications. Neurospora strains containing endogenous locus C-termi- nally VHF tagged PUF4, PRD-2, or untagged negative control were used (Supplementary file 1). Liquid cultures were grown in 2% glucose LCM for 48 hr in constant light. Tissue was harvested by vacuum filtration and fixed by UV cross-linking for 7 min on each side of the fungal mat (Stratalinker UV Crosslinker 1800 with 254 nm wavelength bulbs). UV cross-linked tissue was frozen in liquid nitro- gen and ground into a fine powder with a mortar and pestle. Total protein was extracted in buffer

(25 mM Tris-HCl pH 7.4, 150 mM NaCl, 2 mM MgCl2, 0.5% NP-40, 1 mM DTT, 1Â cOmplete prote- ase inhibitor, 100 U/ml RNAse Out) and concentration determined by Bradford Assay (Bio-Rad # 500–0006). Approximately 10 mg of total protein was added to 30 ml anti-FLAG M2 magnetic beads (Sigma # M8823) prepared according to the manufacturer’s instructions. Beads and lysate were rotated for 4 hr at 4˚C, followed by four washes in 750 ml extraction buffer (5–10 min rotating per wash). Bound RNA-binding proteins were eluted with 100 ml 0.1 M glycine-HCl pH 3.0 for 10 min. The supernatant was collected using a magnetic rack (NEB S1506S) and neutralized in 10 ml of 1 M Tris-HCl pH 8.0. The elution was incubated with 300 ml of TRIzol (Invitrogen # 15596026) for 10 min to extract RNA. Total RNA was isolated, DNAse treated, and concentrated using the Direct-zol RNA Microprep Kit (Zymo # R2062) following the manufacturer’s instructions. Equal amounts of immunoprecipitated RNA (~50 ng) were converted into cDNA using the oligo (dT) method from the SuperScript IV First-Strand synthesis kit (Invitrogen # 18091–050). RT-qPCR was performed using SYBR green master mix (Qiagen # 204054) and a StepOne Plus Real-Time PCR

System (Applied Biosystems). Ct values were determined using StepOne software (Life Technologies)

and normalized to the crp-43 gene (DCt) instead of the actin (NCU04173) gene because actin is a

putative PUF4 target by HITS-CLIP (Wilinski et al., 2017). The DDCt method was used to determine target mRNA enrichment relative to the negative IP sample. Relevant RT-qPCR primer sequences were designed to flank : cbp3 (NCU00057; PUF4 target): 5’ CGAGAAATTCGGCCTTCTCCC 3’ and 5’ GCCTGGTGGAAGAAGTGGT 3’; mrp-1 (NCU07386; PUF4 target): 5’ TAGTAGGCACC- GACTTTGAGCA 3’ and 5’ CGGGGACAGGTGGTCGAA 3’; ck-1a (NCU00685; PRD-2 target): 5’ CGCAAACATGACTACCATG 3’ and 5’ CTCTCCAGCTTGATGGCA 3’; crp-43 (NCU08964; normali- zation control): 5’ CTGTCCGTACTCGTGACTCC 3’ and 5’ ACCATCGATGAGGAGCTTGC 3’.

Protein isolation and detection Frozen Neurospora tissue was ground in liquid nitrogen with a mortar and pestle. Total protein was extracted in buffer (50 mM HEPES pH 7.4, 137 mM NaCl, 10% glycerol v/v, 0.4% NP-40 v/v, and cOmplete Protease Inhibitor Tablet according to instructions for Roche # 11 836 170 001) and proc- essed as described (Garceau et al., 1997). Protein concentrations were determined by Bradford Assay (Bio-Rad # 500–0006). For western blots, equal amounts of total protein (10–30 mg) were loaded per lane into 4–12% Bis-Tris Bolt gels (Invitrogen # NW04125BOX). Western transfer was per- formed using an Invitrogen iBlot system (# IB21001) and PVDF transfer stack (# IB401001). Primary antibodies used for western blotting were anti-V5 (1:3000, ThermoFisher # R960-25), anti-Tubulin alpha (1:10,000, Fitzgerald # 10R-T130a), or anti-CK1a (1:1000, rabbit raised). The secondary anti- bodies, goat anti-mouse or goat anti-rabbit HRP, were used at 1:5000 (Bio-Rad # 170–6516, # 170– 6515). SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher # 34578) or Femto Maximum Sensitivity Substrate (ThermoFisher # 34095) was used for detection. Immunoblot quantifi- cation and normalization were performed in ImageJ. Nuclear and cytosolic fractions were prepared as previously described (Hong et al., 2008). Approximately 10 mg of total protein from each fraction was loaded for immunoblotting. Primary antibodies for fraction controls were histone H3A (Fitzgerald) and g-tubulin (Abcam). HRP-conju- gated secondary antibodies (Bio-Rad) were used with SuperSignal West Pico ECL (Thermo) for detection.

Luciferase reporter detection and data analysis 96-well plates were inoculated with conidial suspensions from strains of interest and entrained in 12 hr light:dark cycles for 2 days in a Percival incubator at 25˚C. Temperature inside the Percival incuba- tor was monitored using a HOBO logger device (Onset # MX2202) during entrainment and free run.

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 16 of 22 Research article Genetics and Genomics Plates were then transferred into constant darkness to initiate the circadian free run. Luminescence was recorded using a Pixis 1024B CCD camera (Princeton Instruments). Light signal was acquired for 10–15 min every hour using LightField software (Princeton Instruments, 64-bit version 6.10.1). The average intensity of each well was determined using a custom ImageJ Macro (Larrondo et al., 2015), and background correction was performed for each frame. Results from two different algo- rithms were averaged together to determine circadian period from background-corrected lumines- cence traces. The MESA algorithm was used as previously described (Kelliher et al., 2020). A second period measurement was obtained from an ordinary least squares autoregressive model to compute the spectral density (in R: spec.ar(..., method=‘ols’)). RT period lengths were measured from scans using ChronOSX 2.1 software (Roenneberg and Taylor, 2000).

Data visualization All figures were plotted in R, output as scalable vector graphics, formatted using Inkscape, and archived in R markdown format. Data represent the mean of at least three biological replicates with standard deviation error bars, unless otherwise indicated.

Acknowledgements We thank the Fungal Genetics Stock Center (Kansas City, Missouri, USA) for curating N. crassa strains. We thank Arun Mehra for discussions on preliminary work to identify the clock-relevant mechanism of the upf1prd-6 mutation, Bin Wang for assistance in constructing and validating the CKISHORT hyperactive allele, Jill Emerson for assistance in constructing Dprd-2 (NCU01019), and Brad Bartholomai for discussions on prd-2. We acknowledge Jerry Feldman for advice on the upf1prd-6 gene naming convention. The Neurospora CK1a antibody was courtesy of Michael Brunner (Univer- sity of Heidelberg). This work was supported by the National Institutes of Health (F32 GM128252 to CMK, R35 GM118021 to JCD, R35 GM118022 to JJL) and EMSL (50173 to Co-PI JCD).

Additional information

Funding Funder Grant reference number Author National Institutes of Health F32 GM128252 Christina M Kelliher National Institutes of Health R35 GM118021 Jay C Dunlap National Institutes of Health R35 GM118022 Jennifer J Loros EMSL 50173 Jay C Dunlap

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Author contributions Christina M Kelliher, Conceptualization, Resources, Data curation, Formal analysis, Funding acquisi- tion, Validation, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing; Randy Lambreghts, Qijun Xiang, Christopher L Baker, Resources, Investigation, Method- ology; Jennifer J Loros, Jay C Dunlap, Conceptualization, Supervision, Funding acquisition, Writing - original draft, Writing - review and editing

Author ORCIDs Christina M Kelliher http://orcid.org/0000-0002-4554-1818 Jay C Dunlap https://orcid.org/0000-0003-1577-0457

Decision letter and Author response Decision letter https://doi.org/10.7554/eLife.64007.sa1 Author response https://doi.org/10.7554/eLife.64007.sa2

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 17 of 22 Research article Genetics and Genomics

Additional files Supplementary files . Supplementary file 1. Neurospora crassa strains used in this study. . Transparent reporting form

Data availability RNA-Sequencing data have been deposited in GEO under accession GSE155999.

The following dataset was generated:

Database and Identifier Author(s) Year Dataset title Dataset URL Kelliher CM, 2020 Nonsense mediated decay https://www.ncbi.nlm. NCBI Gene Expression Lambreghts R, and a novel protein Period-2 nih.gov/geo/query/acc. Omnibus, GSE155999 Xiang Q, Baker CL, regulate casein kinase I in an cgi?acc=GSE155999 Loros JJ, Dunlap JC opposing manner to control circadian period in Neurospora crassa

The following previously published dataset was used:

Database and Author(s) Year Dataset title Dataset URL Identifier Zhang Y, Guo J 2017 RNA-seq analysis of wild type and https://www.ncbi.nlm. NCBI Gene upf1 knockout strains in the nih.gov/geo/query/acc. Expression Omnibus, filamentous fungus Neurospora cgi?acc=GSE97157 GSE97157 crassa

References Aronson BD, Johnson KA, Loros JJ, Dunlap JC. 1994. Negative feedback defining a circadian clock: autoregulation of the clock gene frequency. Science 263:1578–1584. DOI: https://doi.org/10.1126/science. 8128244, PMID: 8128244 Aryal RP, Kwak PB, Tamayo AG, Gebert M, Chiu PL, Walz T, Weitz CJ. 2017. Macromolecular assemblies of the mammalian circadian clock. Molecular Cell 67:770–782. DOI: https://doi.org/10.1016/j.molcel.2017.07.017, PMID: 28886335 Baker CL, Kettenbach AN, Loros JJ, Gerber SA, Dunlap JC. 2009. Quantitative proteomics reveals a dynamic interactome and phase-specific phosphorylation in the Neurospora circadian clock. Molecular Cell 34:354–363. DOI: https://doi.org/10.1016/j.molcel.2009.04.023, PMID: 19450533 Bardiya N, Shiu PK. 2007. Cyclosporin A-resistance based gene placement system for Neurospora crassa. Fungal Genetics and Biology 44:307–314. DOI: https://doi.org/10.1016/j.fgb.2006.12.011, PMID: 17320431 Belden WJ, Larrondo LF, Froehlich AC, Shi M, Chen CH, Loros JJ, Dunlap JC. 2007. The band mutation in Neurospora crassa is a dominant allele of ras-1 implicating RAS signaling in circadian output. Genes & Development 21:1494–1505. DOI: https://doi.org/10.1101/gad.1551707, PMID: 17575051 Chen CH, Ringelberg CS, Gross RH, Dunlap JC, Loros JJ. 2009. Genome-wide analysis of light-inducible responses reveals hierarchical light signalling in Neurospora. The EMBO Journal 28:1029–1042. DOI: https:// doi.org/10.1038/emboj.2009.54, PMID: 19262566 Cheong JK, Virshup DM. 2011. : complexity in the family. The International Journal of Biochemistry & Cell Biology 43:465–469. DOI: https://doi.org/10.1016/j.biocel.2010.12.004, PMID: 21145983 Colot HV, Loros JJ, Dunlap JC. 2005. Temperature-modulated alternative splicing and promoter use in the circadian clock gene frequency. Molecular Biology of the Cell 16:5563–5571. DOI: https://doi.org/10.1091/ mbc.e05-08-0756, PMID: 16195340 Compton J. 2003. Advances in understanding the molecular biology of Neurospora crassa. University of California Santa Cruz, Ph.D. Thesis. Diernfellner AC, Schafmeier T, Merrow MW, Brunner M. 2005. Molecular mechanism of temperature sensing by the circadian clock of Neurospora crassa. Genes & Development 19:1968–1973. DOI: https://doi.org/10.1101/ gad.345905, PMID: 16107616 Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, Batut P, Chaisson M, Gingeras TR. 2013. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29:15–21. DOI: https://doi.org/10.1093/bioinformatics/ bts635, PMID: 23104886 Dunlap JC, Loros JJ. 2018. Just-So stories and origin myths: phosphorylation and structural disorder in circadian clock proteins. Molecular Cell 69:165–168. DOI: https://doi.org/10.1016/j.molcel.2017.11.028, PMID: 29276084

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 18 of 22 Research article Genetics and Genomics

Emerson JM, Bartholomai BM, Ringelberg CS, Baker SE, Loros JJ, Dunlap JC. 2015. period-1 encodes an ATP- dependent RNA helicase that influences nutritional compensation of the Neurospora circadian clock. PNAS 112:15707–15712. DOI: https://doi.org/10.1073/pnas.1521918112, PMID: 26647184 Feldman JF, Hoyle MN. 1973. Isolation of circadian clock mutants of Neurospora crassa. Genetics 75:605–613. PMID: 4273217 Feldman JF, Hoyle MN. 1976. Complementation analysis of linked circadian clock mutants of Neurospora crassa. Genetics 82:9–17. PMID: 129346 Fustin JM, Kojima R, Itoh K, Chang HY, Ye S, Zhuang B, Oji A, Gibo S, Narasimamurthy R, Virshup D, Kurosawa G, Doi M, Manabe I, Ishihama Y, Ikawa M, Okamura H. 2018. Two Ck1d transcripts regulated by m6A methylation code for two antagonistic kinases in the control of the circadian clock. PNAS 115:5980–5985. DOI: https://doi.org/10.1073/pnas.1721371115, PMID: 29784786 Garceau NY, Liu Y, Loros JJ, Dunlap JC. 1997. Alternative initiation of translation and time-specific phosphorylation yield multiple forms of the essential clock protein FREQUENCY. Cell 89:469–476. DOI: https:// doi.org/10.1016/S0092-8674(00)80227-5, PMID: 9150146 Gardner GF, Feldman JF. 1981. Temperature compensation of circadian period length in clock mutants of Neurospora crassa. Plant Physiology 68:1244–1248. DOI: https://doi.org/10.1104/pp.68.6.1244, PMID: 166620 86 Ge Z, Quek BL, Beemon KL, Hogg JR. 2016. Polypyrimidine tract binding protein 1 protects mRNAs from recognition by the nonsense-mediated mRNA decay pathway. eLife 5:e11155. DOI: https://doi.org/10.7554/ eLife.11155, PMID: 26744779 Gietzen KF, Virshup DM. 1999. Identification of inhibitory autophosphorylation sites in casein kinase I e. Journal of Biological Chemistry 274:32063–32070. DOI: https://doi.org/10.1074/jbc.274.45.32063 Gooch VD, Mehra A, Larrondo LF, Fox J, Touroutoutoudis M, Loros JJ, Dunlap JC. 2008. Fully codon-optimized luciferase uncovers novel temperature characteristics of the Neurospora clock. Eukaryotic Cell 7:28–37. DOI: https://doi.org/10.1128/EC.00257-07, PMID: 17766461 Go¨ rl M, Merrow M, Huttner B, Johnson J, Roenneberg T, Brunner M. 2001. A PEST-like element in FREQUENCY determines the length of the circadian period in Neurospora crassa. The EMBO Journal 20:7074–7084. DOI: https://doi.org/10.1093/emboj/20.24.7074, PMID: 11742984 Guo J, Cheng P, Yuan H, Liu Y. 2009. The exosome regulates circadian gene expression in a posttranscriptional negative feedback loop. Cell 138:1236–1246. DOI: https://doi.org/10.1016/j.cell.2009.06.043, PMID: 19747717 Guo G, Wang K, Hu SS, Tian T, Liu P, Mori T, Chen P, Johnson CH, Qin X. 2019. Autokinase activity of casein kinase 1 d/e governs the period of mammalian circadian rhythms. Journal of Biological Rhythms 34:482–496. DOI: https://doi.org/10.1177/0748730419865406, PMID: 31392916 He Q, Cha J, He Q, Lee HC, Yang Y, Liu Y. 2006. CKI and CKII mediate the FREQUENCY-dependent phosphorylation of the WHITE COLLAR complex to close the Neurospora circadian negative feedback loop. Genes & Development 20:2552–2565. DOI: https://doi.org/10.1101/gad.1463506, PMID: 16980584 Heintzen C, Loros JJ, Dunlap JC. 2001. The PAS protein VIVID defines a clock-associated feedback loop that represses light input, modulates gating, and regulates clock resetting. Cell 104:453–464. DOI: https://doi.org/ 10.1016/S0092-8674(01)00232-X, PMID: 11239402 Hong CI, Ruoff P, Loros JJ, Dunlap JC. 2008. Closing the circadian negative feedback loop: FRQ-dependent clearance of WC-1 from the nucleus. Genes & Development 22:3196–3204. DOI: https://doi.org/10.1101/gad. 1706908, PMID: 18997062 Hurley JM, Larrondo LF, Loros JJ, Dunlap JC. 2013. Conserved RNA helicase FRH acts nonenzymatically to support the intrinsically disordered Neurospora clock protein FRQ. Molecular Cell 52:832–843. DOI: https:// doi.org/10.1016/j.molcel.2013.11.005, PMID: 24316221 Hurley JM, Dasgupta A, Emerson JM, Zhou X, Ringelberg CS, Knabe N, Lipzen AM, Lindquist EA, Daum CG, Barry KW, Grigoriev IV, Smith KM, Galagan JE, Bell-Pedersen D, Freitag M, Cheng C, Loros JJ, Dunlap JC. 2014. Analysis of clock-regulated genes in Neurospora reveals widespread posttranscriptional control of metabolic potential. PNAS 111:16995–17002. DOI: https://doi.org/10.1073/pnas.1418963111, PMID: 25362047 Hurley JM, Loros JJ, Dunlap JC. 2016. Circadian oscillators: around the Transcription-Translation feedback loop and on to output. Trends in Biochemical Sciences 41:834–846. DOI: https://doi.org/10.1016/j.tibs.2016.07.009, PMID: 27498225 Hurley JM, Jankowski MS, De Los Santos H, Crowell AM, Fordyce SB, Zucker JD, Kumar N, Purvine SO, Robinson EW, Shukla A, Zink E, Cannon WR, Baker SE, Loros JJ, Dunlap JC. 2018. Circadian proteomic analysis uncovers mechanisms of Post-Transcriptional regulation in metabolic pathways. Cell Systems 7:613–626. DOI: https://doi.org/10.1016/j.cels.2018.10.014, PMID: 30553726 Isojima Y, Nakajima M, Ukai H, Fujishima H, Yamada RG, Masumoto KH, Kiuchi R, Ishida M, Ukai-Tadenuma M, Minami Y, Kito R, Nakao K, Kishimoto W, Yoo SH, Shimomura K, Takao T, Takano A, Kojima T, Nagai K, Sakaki Y, et al. 2009. CKIepsilon/delta-dependent phosphorylation is a temperature-insensitive, period-determining process in the mammalian circadian clock. PNAS 106:15744–15749. DOI: https://doi.org/10.1073/pnas. 0908733106, PMID: 19805222 James AB, Syed NH, Bordage S, Marshall J, Nimmo GA, Jenkins GI, Herzyk P, Brown JW, Nimmo HG. 2012. Alternative splicing mediates responses of the Arabidopsis circadian clock to temperature changes. The Plant Cell 24:961–981. DOI: https://doi.org/10.1105/tpc.111.093948, PMID: 22408072 Kelliher CM, Loros JJ, Dunlap JC. 2020. Evaluating the and response to glucose addition in dispersed growth cultures of Neurospora crassa. Fungal Biology 124:398–406. DOI: https://doi.org/10.1016/j. funbio.2019.11.004, PMID: 32389302

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 19 of 22 Research article Genetics and Genomics

Kloss B, Rothenfluh A, Young MW, Saez L. 2001. Phosphorylation of period is influenced by cycling physical associations of double-time, period, and in the Drosophila clock. Neuron 30:699–706. DOI: https://doi. org/10.1016/S0896-6273(01)00320-8, PMID: 11430804 Koike N, Yoo SH, Huang HC, Kumar V, Lee C, Kim TK, Takahashi JS. 2012. Transcriptional architecture and chromatin landscape of the core circadian clock in mammals. Science 338:349–354. DOI: https://doi.org/10. 1126/science.1226339, PMID: 22936566 Konopka RJ, Benzer S. 1971. Clock mutants of . PNAS 68:2112–2116. DOI: https://doi. org/10.1073/pnas.68.9.2112 Kramer C, Loros JJ, Dunlap JC, Crosthwaite SK. 2003. Role for antisense RNA in regulating circadian clock function in Neurospora crassa. Nature 421:948–952. DOI: https://doi.org/10.1038/nature01427 Lambreghts R. 2012. Exploring New Players in the Neurospora Core Clock and its Output. Dartmouth College, Ph.D. Thesis. Larrondo LF, Olivares-Yan˜ ez C, Baker CL, Loros JJ, Dunlap JC. 2015. Circadian rhythms decoupling circadian clock protein turnover from circadian period determination. Science 347:1257277. DOI: https://doi.org/10. 1126/science.1257277, PMID: 25635104 Lauinger L, Li J, Shostak A, Cemel IA, Ha N, Zhang Y, Merkl PE, Obermeyer S, Stankovic-Valentin N, Schafmeier T, Wever WJ, Bowers AA, Carter KP, Palmer AE, Tschochner H, Melchior F, Deshaies RJ, Brunner M, Diernfellner A. 2017. Thiolutin is a zinc Chelator that inhibits the Rpn11 and other JAMM metalloproteases. Nature Chemical Biology 13:709–714. DOI: https://doi.org/10.1038/nchembio.2370, PMID: 28459440 Lee H, Chen R, Lee Y, Yoo S, Lee C. 2009. Essential roles of CKIdelta and CKIepsilon in the mammalian circadian clock. PNAS 106:21359–21364. DOI: https://doi.org/10.1073/pnas.0906651106, PMID: 19948962 Lewis M. 1995. Molecular genetic analysis of circadian clock genes in Neurospora crassa. University of California Santa Cruz, Ph.D. Thesis. Liu Y, Garceau NY, Loros JJ, Dunlap JC. 1997. Thermally regulated translational control of FRQ mediates aspects of temperature responses in the Neurospora circadian clock. Cell 89:477–486. DOI: https://doi.org/10.1016/ S0092-8674(00)80228-7, PMID: 9150147 Liu X, Chen A, Caicedo-Casso A, Cui G, Du M, He Q, Lim S, Kim HJ, Hong CI, Liu Y. 2019. FRQ-CK1 interaction determines the period of circadian rhythms in Neurospora. Nature Communications 10:1–13. DOI: https://doi. org/10.1038/s41467-019-12239-w Loros JJ. 2020. Principles of the animal molecular clock learned from Neurospora. European Journal of Neuroscience 51:19–33. DOI: https://doi.org/10.1111/ejn.14354 Majercak J, Sidote D, Hardin PE, Edery I. 1999. How a circadian clock adapts to seasonal decreases in temperature and day length. Neuron 24:219–230. DOI: https://doi.org/10.1016/S0896-6273(00)80834-X, PMID: 10677039 Mateos JL, de Leone MJ, Torchio J, Reichel M, Staiger D. 2018. Beyond transcription: fine-tuning of circadian timekeeping by Post-Transcriptional regulation. Genes 9:616. DOI: https://doi.org/10.3390/genes9120616 Mehra A, Shi M, Baker CL, Colot HV, Loros JJ, Dunlap JC. 2009. A role for in the mechanism underlying circadian temperature compensation. Cell 137:749–760. DOI: https://doi.org/10.1016/j.cell.2009.03. 019, PMID: 19450520 Metzenberg RL. 2004. Bird medium: an alternative to vogel medium. Fungal Genetics Reports 51:19–20. DOI: https://doi.org/10.4148/1941-4765.1138 Millius A, Ueda HR. 2017. Systems Biology-Derived discoveries of intrinsic clocks. Frontiers in Neurology 8:1–19. DOI: https://doi.org/10.3389/fneur.2017.00025 Morgan LW, Feldman JF. 1997. Isolation and characterization of a temperature-sensitive circadian clock mutant of Neurospora crassa. Genetics 146:525–530. PMID: 9178003 Morgan LW, Feldman JF. 2001. Epistatic and synergistic interactions between circadian clock mutations in Neurospora crassa. Genetics 159:537–543. PMID: 11606531 Nakashima H. 1981. A liquid culture method for the biochemical analysis of the circadian clock of Neurospora crassa. Plant & Cell Physiology 22:231–238. DOI: https://doi.org/10.1093/oxfordjournals.pcp.a076160 Narasimamurthy R, Hunt SR, Lu Y, Fustin JM, Okamura H, Partch CL, Forger DB, Kim JK, Virshup DM. 2018. CK1d/e protein kinase primes the PER2 circadian phosphoswitch. PNAS 115:5986–5991. DOI: https://doi.org/ 10.1073/pnas.1721076115, PMID: 29784789 Ode KL, Ukai H, Susaki EA, Narumi R, Matsumoto K, Hara J, Koide N, Abe T, Kanemaki MT, Kiyonari H, Ueda HR. 2017. Knockout-Rescue embryonic stem Cell-Derived mouse reveals Circadian-Period control by quality and quantity of CRY1. Molecular Cell 65:176–190. DOI: https://doi.org/10.1016/j.molcel.2016.11.022, PMID: 2 8017587 Partch CL. 2020. Orchestration of circadian timing by macromolecular protein assemblies. Journal of Molecular Biology 432:3426–3448. DOI: https://doi.org/10.1016/j.jmb.2019.12.046, PMID: 31945377 Pelham JF, Mosier AE, Hurley JM. 2018. Characterizing Time-of-Day conformational changes in the intrinsically disordered proteins of the circadian clock. Methods in Enzymology 611:503–529. DOI: https://doi.org/10.1016/ bs.mie.2018.08.024, PMID: 30471697 Pelham JF, Dunlap JC, Hurley JM. 2020. Intrinsic disorder is an essential characteristic of components in the conserved circadian circuit. Cell Communication and Signaling 18:181. DOI: https://doi.org/10.1186/s12964- 020-00658-y, PMID: 33176800 Philpott JM, Narasimamurthy R, Ricci CG, Freeberg AM, Hunt SR, Yee LE, Pelofsky RS, Tripathi S, Virshup DM, Partch CL. 2020. Casein kinase 1 dynamics underlie substrate selectivity and the PER2 circadian phosphoswitch. eLife 9:e52343. DOI: https://doi.org/10.7554/eLife.52343, PMID: 32043967

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 20 of 22 Research article Genetics and Genomics

Pregueiro AM, Liu Q, Baker CL, Dunlap JC, Loros JJ. 2006. The Neurospora checkpoint kinase 2: a regulatory link between the circadian and cell cycles. Science 313:644–649. DOI: https://doi.org/10.1126/science. 1121716, PMID: 16809488 Querfurth C, Diernfellner A, Heise F, Lauinger L, Neiss A, Tataroglu O¨ ., Brunner M, Schafmeier T. 2007. Posttranslational regulation of Neurospora circadian clock by CK1a-dependent phosphorylation. Cold Spring Harbor Symposia on Quantitative Biology 177–183. Ralph M, Menaker M. 1988. A mutation of the circadian system in golden hamsters. Science 241:1225–1227. DOI: https://doi.org/10.1126/science.3413487 Ri H, Lee J, Sonn JY, Yoo E, Lim C, Choe J. 2019. Drosophila CrebB is a substrate of the Nonsense-Mediated mRNA decay pathway that sustains circadian behaviors. Molecules and Cells 42:301–312. DOI: https://doi.org/ 10.14348/molcells.2019.2451, PMID: 31091556 Roenneberg T, Taylor W. 2000. Automated recordings of bioluminescence with special reference to the analysis of circadian rhythms. Methods in Enzymology 305:104–119. DOI: https://doi.org/10.1016/s0076-6879(00) 05481-1, PMID: 10812594 Sargent ML, Briggs WR, Woodward DO. 1966. Circadian nature of a rhythm expressed by an invertaseless strain of Neurospora crassa. Plant Physiology 41:1343–1349. DOI: https://doi.org/10.1104/pp.41.8.1343, PMID: 597 8549 Song MH, Aravind L, Mu¨ ller-Reichert T, O’Connell KF. 2008. The conserved protein SZY-20 opposes the Plk4- related kinase ZYG-1 to limit centrosome size. Developmental Cell 15:901–912. DOI: https://doi.org/10.1016/j. devcel.2008.09.018, PMID: 19081077 Sundaram S, Nagaraj S, Mahoney H, Portugues A, Li W, Millsaps K, Faulkner J, Yunus A, Burns C, Bloom C, Said M, Pinto L, Azam S, Flores M, Henriksen A, Gamsby J, Gulick D. 2019. Inhibition of casein kinase 1d/e improves cognitive-affective behavior and reduces amyloid load in the APP-PS1 mouse model of Alzheimer’s disease. Scientific Reports 9:1–13. DOI: https://doi.org/10.1038/s41598-019-50197-x Tan Y, Dragovic Z, Roenneberg T, Merrow M. 2004. Entrainment dissociates transcription and translation of a circadian clock gene in Neurospora. Current Biology 14:433–438. DOI: https://doi.org/10.1016/j.cub.2004.02. 035, PMID: 15028220 Toh KL, Jones CR, He Y, Eide EJ, Hinz WA, Virshup DM, Pta´cek LJ, Fu YH. 2001. An hPer2 phosphorylation site mutation in familial advanced sleep phase syndrome. Science 291:1040–1043. DOI: https://doi.org/10.1126/ science.1057499, PMID: 11232563 Top D, O’Neil JL, Merz GE, Dusad K, Crane BR, Young MW. 2018. CK1/Doubletime activity delays transcription activation in the circadian clock. eLife 7:e32679. DOI: https://doi.org/10.7554/eLife.32679, PMID: 29611807 Trapnell C, Hendrickson DG, Sauvageau M, Goff L, Rinn JL, Pachter L. 2013. Differential analysis of gene regulation at transcript resolution with RNA-seq. Nature Biotechnology 31:46–53. DOI: https://doi.org/10. 1038/nbt.2450, PMID: 23222703 Tsuchiya Y, Umemura Y, Minami Y, Koike N, Hosokawa T, Hara M, Ito H, Inokawa H, Yagita K. 2016. Effect of multiple clock gene ablations on the circadian period length and temperature compensation in mammalian cells. Journal of Biological Rhythms 31:48–56. DOI: https://doi.org/10.1177/0748730415613888, PMID: 26511603 Vanselow K, Vanselow JT, Westermark PO, Reischl S, Maier B, Korte T, Herrmann A, Herzel H, Schlosser A, Kramer A. 2006. Differential effects of PER2 phosphorylation: molecular basis for the human familial advanced sleep phase syndrome (FASPS). Genes & Development 20:2660–2672. DOI: https://doi.org/10.1101/gad. 397006, PMID: 16983144 Vielhaber E, Virshup DM. 2001. Casein kinase I: from obscurity to center stage. IUBMB Life 51:73–78. DOI: https://doi.org/10.1080/15216540152122049, PMID: 11463166 Wang B, Kettenbach AN, Zhou X, Loros JJ, Dunlap JC. 2019. The Phospho-Code determining circadian feedback loop closure and output in Neurospora. Molecular Cell 74:771–784. DOI: https://doi.org/10.1016/j.molcel. 2019.03.003, PMID: 30954403 Weischenfeldt J, Waage J, Tian G, Zhao J, Damgaard I, Jakobsen JS, Kristiansen K, Krogh A, Wang J, Porse BT. 2012. Mammalian tissues defective in nonsense-mediated mRNA decay display highly aberrant splicing patterns. Genome Biology 13:R35. DOI: https://doi.org/10.1186/gb-2012-13-5-r35, PMID: 22624609 Wilinski D, Buter N, Klocko AD, Lapointe CP, Selker EU, Gasch AP, Wickens M. 2017. Recurrent rewiring and emergence of RNA regulatory networks. PNAS 114:E2816–E2825. DOI: https://doi.org/10.1073/pnas. 1617777114, PMID: 28320951 Wu C, Yang F, Smith KM, Peterson M, Dekhang R, Zhang Y, Zucker J, Bredeweg EL, Mallappa C, Zhou X, Lyubetskaya A, Townsend JP, Galagan JE, Freitag M, Dunlap JC, Bell-Pedersen D, Sachs MS. 2014. Genome- Wide characterization of Light-Regulated genes in Neurospora crassa. G3: Genes, Genomes, Genetics 4:1731– 1745. DOI: https://doi.org/10.1534/g3.114.012617 Wu Y, Zhang Y, Sun Y, Yu J, Wang P, Ma H, Chen S, Ma L, Zhang D, He Q, Guo J. 2017. Up-Frameshift protein UPF1 regulates Neurospora crassa Circadian and Diurnal Growth Rhythms. Genetics 206:1881–1893. DOI: https://doi.org/10.1534/genetics.117.202788, PMID: 28600326 Xu Y, Padiath QS, Shapiro RE, Jones CR, Wu SC, Saigoh N, Saigoh K, Pta´cek LJ, Fu YH. 2005. Functional consequences of a CKIdelta mutation causing familial advanced sleep phase syndrome. Nature 434:640–644. DOI: https://doi.org/10.1038/nature03453, PMID: 15800623 Yoo S-H, Kojima S, Shimomura K, Koike N, Buhr ED, Furukawa T, Ko CH, Gloston G, Ayoub C, Nohara K, Reyes BA, Tsuchiya Y, Yoo O-J, Yagita K, Lee C, Chen Z, Yamazaki S, Green CB, Takahashi JS. 2017. Period2 30-UTR

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 21 of 22 Research article Genetics and Genomics

and microRNA-24 regulate circadian rhythms by repressing PERIOD2 protein accumulation. PNAS 114:E8855– E8864. DOI: https://doi.org/10.1073/pnas.1706611114 Zhang Y, Sachs MS. 2015. Control of mRNA stability in fungi by NMD, EJC and CBC factors through 3’UTR Introns. Genetics 200:1133–1148. DOI: https://doi.org/10.1534/genetics.115.176743, PMID: 26048019 Zhou M, Guo J, Cha J, Chae M, Chen S, Barral JM, Sachs MS, Liu Y. 2013. Non-optimal Codon usage affects expression, structure and function of clock protein FRQ. Nature 495:111–115. DOI: https://doi.org/10.1038/ nature11833, PMID: 23417067 Zhou M, Kim JK, Eng GW, Forger DB, Virshup DM. 2015. A Period2 phosphoswitch regulates and temperature compensates circadian period. Molecular Cell 60:77–88. DOI: https://doi.org/10.1016/j.molcel.2015.08.022, PMID: 26431025

Kelliher et al. eLife 2020;9:e64007. DOI: https://doi.org/10.7554/eLife.64007 22 of 22